Abstract: With the deepening of the study of power quality, the problems leading by voltage sag have received more attention. As the applications based on sensitive devices such as microcomputers, digital-control drives, variable-frequency drives and so on, become more widespread, the problems leading by voltage sag are also increasing. For the users in power system, every time the voltage sag event happens, it is accompanied by huge economic losses. In this paper, an integrated method of quality loss functions based on the entropy weight method is proposed to evaluate the economic losses of users. For a power user, the economic losses caused by voltage sag include the losses of waste, the costs of remediation of production, the costs of equipment, the costs of lost profits and restart, and so on. Some of the losses are due to disruptions of productive activities, while the other are the economic losses that would exist even if the production was not interrupted. In this paper, we consider these two kinds of losses separately. The losses caused by disruptions of productive activities are the losses due to the equipment that quits work because of outage. Each power user may have more than one kind of device that is sensitive to voltage sag, also due to different factors such as different manufacturers, the voltage-tolerance curve of sensitive device is different. According to different equipment of power users, we can find a comprehensive voltage-tolerance curve to reflect the voltage-tolerance ability of all user’s equipment. A probabilistic model of utility outage considering energy loss is used to calculate the probability of utility outage. Then the quality loss function is introduced to describe the relationship between utility outage and the economic losses caused by disruptions of production activities. For the economic losses that exist without disruptions of productive activities, we need find the relationship between the characteristics of voltage sag and this kind of economic losses. According to the indices of IEEE Std 1564-2014 for single-event, we use them as independent variables of the quality loss of inverse normal function, and combine the entropy weight method to sum up multi-class quality loss functions. Finally, a power grid model is taken as a simulation example, the data of economic losses of power users on one past year is used as a reference. The simulation results show that this method can effectively obtain the economic losses of voltage sags, and provide useful theoretical guidance for the assessment, and improve power quality.